January 2022 Summaries

Deploy Authorizer on Railway` is an open-source authentication and authorization solution that provides a self-hosted alternative to Auth0. The solution comes with easy-to-integrate frontend SDKs and allows users to bring their own database. Railway offers a free one-click deploy for Authorizer, which is maintained by its creator Lakhan Samani, along with automatically provisioned PostgreSQL and Redis databases. To deploy the solution, users can click a button on Railway's platform and follow a simple guide, providing features such as email and password-based accounts, social logins, secure session management, role-based access management, and more. The deployment process involves creating a repository in the user's GitHub account and filling in required environment variables before hitting the deploy button to launch their new project with an Authorizer instance deployed in a few minutes. Additionally, users can integrate Authorizer into their React frontend applications using a provided hook and component.

We at Railway use a unique approach to product development by integrating our team chats with user help chats, allowing for immediate collaboration and feedback. Our Discord integration enables users to link their accounts, granting access to pre-release features and providing a tight feedback loop. We've implemented role caching to efficiently check user roles, and Percy, our Discord bot, helps manage the beta program, assigning users to the `@priority-boarding` role and enabling them to request access. This approach has led to invaluable user feedback, influencing product design decisions and helping us build better products.

We've rebuilt our log storage and search infrastructure for scale by replacing a polling-based system with a more scalable solution that uses Docker's syslog driver, Vector log aggregator, Google Cloud Storage, and Go micro-service using GRPC streams. The new system allows users to view logs for all historical deployments and filter logs by keyword in real-time, making it more efficient and scalable than the previous implementation.